Another great question asked in Ask Away Round 3 was about statistics and when to know to pause your marketing campaigns or to continue testing.

“Right now I’m trying out FB ads, and I don’t know when ads have gotten enough impressions to consider the CTR significant. Or how many clicks I need until the conversion rate for a particular ad would be considered significant.

How do you know when your information is significant? Especially when the samples size for the facebook demo vary, for example if you have a group of 5000 for one campaign and 30000 for another. Would you need less data for the first group?”

Number 1, I’ll tell you straight up that CTR on your Facebook ad isn’t as important as your Cost Per Conversion. You can have an ad with a 4% CTR on Facebook, but it might only convert once in one hundred clicks. Alternatively, you might write another ad that gets a 1% CTR but a 20% CVR. I don’t need to tell you which one is better there. My buddy, Khoa wrote a blog post about this. Check it out.

Overall, focus on raising your CVR/Cost per conversion rather than focusing mainly on ad CTR.

If you are bidding CPM, which you honestly should be by now, I normally allocate a budget of 3 – 5 times the offer payout to test. If I don’t get 1 conversion spread over a few different ad types, I’ll either re-target or trash the campaign/offer all together. At that point, you either have to realize you aren’t targeting/selling correctly, or it just isn’t going to work out for you.

With smaller demographic targets, you are going to have less leeway for testing, however you won’t have to spend that much testing it out either. If you’ve got a small target of 5000 people, number 1, cycle through a lot of different ads (Use Facebook Ad Manager to quickly and automagically build multiple ad variations), rotate constantly. Banner blindness sets in pretty fast when you are bidding high for a smaller domain of people. If you can’t profit off them in the first pass, but you see  a way that you could possibly start to profit (say you are only running -20% ROI), try a second pass with new ads and an optimized landing page. Trash it otherwise. If you see a profit, constantly change up ads. After a week or two you will probably dry up the target range, so pause for 2-4 weeks and fire it up again. This seems to have worked for me when I’d have targets of 100 – 1000 people.

Statistics is a funny game, because you can average out everything, and get a good estimated result frame from what you are doing. However, that isn’t always the result. I think you can still get  a lot out of basic intuition, common sense, as well as being in the customers mindset when viewing an ad campaign you are about to unleash.

Another commenter left this A/B split test link in the comments section. This can help you A) figure out which ad groups convert the best and B) whether or not those conversion rates are of a high confidence level. I’m really no mathematician so a lot of my data I just look at what is giving me the absolute highest amount of money for my costs and I pretty much believe that is what any affiliate marketer should look at.